AI Periodic Table — Engram Capability Matrix

Engram’s competitive position mapped to Martin Keen’s AI Periodic Table framework.

Video Introduction

*Credit: Martin Keen, IBM Master Inventor Video*

Interactive Matrix

📊 View Full Interactive Matrix

Status Legend

Status Meaning Count
🟢 Strong Production-ready implementation 12
🟡 Emerging Partial implementation or roadmap 1
🔴 Gap Not yet implemented 3
⭐ Unique Competitive differentiator 1

Element-by-Element Breakdown

Row 1: Primitives

Element Symbol Status Engram Component Dashboard
Prompts Pr 🟢 Strong Agent system prompts Agent Config
Embeddings Em 🟢 Strong Zep embedding layer Zep Cloud
LLM Lg 🟢 Strong Claude, Gemini, GPT-4o Azure AI

Row 2: Compositions

Element Symbol Status Engram Component Dashboard
Function Call Fc 🟢 Strong Story Gen delegate_to_sage
GH Issues create_github_issue
OneDrive save_to_onedrive		 MCP Docs
Vector Vx 🟢 Strong Zep Tri-Search (Vector Layer) Zep Cloud
RAG Rg 🟢 Strong Context assembly pipeline (Tri-Search Fusion) Architecture
Guardrails Gr 🟢 Strong Azure Entra ID Entra Admin
Multimodal Mm 🟢 Strong Imagen 3.0 + VoiceLive Azure AI

Row 3: Deployment

Element Symbol Status Engram Component Dashboard
Agent Ag 🟢 Strong Elena, Marcus, Sage Agent Config
Finetune Ft 🔴 Gap Roadmap: Granite/Llama —
Framework Fw 🟢 Strong Temporal workflows Temporal UI
Red-team Rt 🟡 Emerging Basic validation —
Small Sm 🔴 Gap Roadmap: Phi-4/Granite —

Row 4: Emerging

Element Symbol Status Engram Component Dashboard
Multi-agent Ma 🟢 Strong Agent delegation Workflow Monitor
Synthetic Sy 🟢 Strong Story + visual gen Stories
Graph Knowledge Gk ⭐ Unique Zep Temporal KG + Visualization Graph Interface
Interpret In 🔴 Gap Roadmap: Explainability —
Thinking Th 🟢 Strong Extended reasoning Workflow Monitor

Gk (Graph Knowledge) — Our Unique Advantage

Symbol: Gk
Position: Row 4 (Emerging) × Column 3 (Orchestration)
Definition: Temporal knowledge graphs for dynamic context orchestration
Status: ⭐ Unique Differentiator — Production-ready with enhanced visualization

Why Gk Matters

While competitors use static RAG (retrieve → augment → generate), Engram uses dynamic Graph Knowledge orchestration:

Aspect Static Frameworks (Fw) Graph Knowledge (Gk)
Routing Logic Predefined chains/DAGs Dynamic, semantic routing
Context Assembly Manual prompt engineering Automatic via graph traversal
Memory Stateless or session-scoped Temporal, multi-session knowledge
Discovery Explicit tool registration Emergent via entity relationships

Tri-Search: The Complete Picture

Graph Knowledge is the critical third layer of Engram’s tri-search capability:

  1. Keyword Search (BM25): Exact phrase matching, acronym lookup
  2. Vector Search (Semantic): Conceptual similarity via embeddings
  3. Graph Search (Gk): Relationship traversal, multi-hop reasoning ⭐

Results from all three layers are combined using Reciprocal Rank Fusion (RRF) for optimal retrieval.

📖 Full Documentation: Graph Knowledge & Tri-Search Guide

Implementation

Engram implements Gk through Zep Cloud:

  • Entity Extraction: Automatic extraction from conversations and documents
  • Fact Linking: Relationships stored as graph edges with timestamps
  • Temporal Awareness: Facts have timestamps, enabling time-aware queries
  • Cross-Session Learning: Knowledge compounds automatically across conversations
  • Visualization: Interactive graph interface at /memory/graph

Graph Knowledge Interface

Access: Knowledge Graph Visualization

Features:

  • Interactive Force-Directed Graph: Visualize relationships between entities, facts, episodes, and topics
  • Search & Filter: Query-based filtering, node type filters, degree-based filtering
  • Statistics Dashboard: Total nodes/edges, average degree, node type breakdown
  • Node Details: Full content, metadata, connections, and traversal paths
  • Tri-Search Context: Explanation of how Graph Knowledge fits into tri-search

Node Types:

  • Facts (Cyan): Semantic facts extracted from conversations
  • Entities (Purple): People, projects, concepts
  • Episodes (Green): Conversation sessions/episodic memory
  • Topics (Amber): Conversation themes and topics
  • Metadata (Gray): Source tags, filenames, tenant IDs

Use Cases

  1. Entity Discovery: “Who worked on the authentication project?” → Traverse graph to find connected people
  2. Project Timeline: “What happened with Project Alpha over time?” → Chronological episode traversal
  3. Knowledge Gap Analysis: Identify isolated nodes (low degree) that need more context
  4. Multi-Hop Reasoning: “What did Elena say about topics related to security?” → Multi-step graph traversal

Observability

What You Can See:

  • ✅ Graph structure and relationships
  • ✅ Node details (content, metadata, connections)
  • ✅ Statistics (total nodes, edges, degree metrics)
  • ✅ Search transparency (which nodes match query)

Roadmap:

  • 🔄 Tri-search breakdown (which layer contributed each result)
  • 🔄 Retrieval path visualization
  • 🔄 Graph analytics (centrality, community detection)
  • 🔄 Time slider (view graph at different time points)

Dashboards & Observability

System Dashboard Purpose
Azure AI Portal Model deployments, costs
Temporal UI Workflow monitoring
Zep Cloud Memory, embeddings, graph
Knowledge Graph Interface Graph visualization, tri-search layer 3
Entra ID Admin Authentication, RBAC
Cost Management Azure FinOps

Provenance